论文部分内容阅读
球磨机是应用于磨矿过程的关键设备,可靠测量其料位对实现优化控制和节能降耗具有重大意义。针对球磨机振声信号存在非线性及其梅尔频率倒谱系数(MFCC)存在不相关信息等问题,提出了一种基于MFCC、受限玻尔兹曼机(RBM)和支持向量回归(SVR)的软测量方法。首先对振声信号计算MFCC特征,然后采用RBM对MFCC进行特征提取,以减弱噪声和提高模型精度,最后将提取的特征输入支持向量机进行回归。实验结果表明该模型的有效性和可行性。
Ball mill is the key equipment applied in the grinding process. Reliable measurement of the material level is of great significance to achieve optimal control and energy saving. In view of the non-linearity of vibration signal of ball mill and the irrelevant information of its Mel Frequency Cepstral Coefficient (MFCC), this paper proposes a new algorithm based on MFCC, Restricted Boltzmann Machine (RBM) and Support Vector Regression (SVR) Soft measurement method. Firstly, the MFCC feature is calculated for the vibration signal, and then the feature extraction is performed on the MFCC by using RBM to reduce the noise and improve the accuracy of the model. Finally, the extracted feature is input into the support vector machine for regression. Experimental results show that the model is effective and feasible.